1# WP35 — Pre-registration of the empirical tests23**Status: FROZEN on 2026-11-01 before any predictor–outcome association was computed.** The SHA-256 of this file at4freeze time is recorded in `research/decisions.md` (D-003). Any later change is an amendment, numbered in5`research/decisions.md`, with its date, reason, and whether it was made before or after seeing the corresponding results.67Author of the plan: the autonomous executor (Claude Code), under the PI's mandate (`docs/MANDATE_VERBATIM.txt`).8What was inspected before writing this plan (and nothing else): the structure of every raw file (sheet names, variable9names, row counts, value vocabularies, date coverage, outcome code frequencies) — see `results/logs/feasibility_report.md`10and the execution log. **No cross-tabulation or correlation between any candidate predictor and any outcome was11examined.**1213---1415## 1. Research question and hypotheses as they will be tested1617The theoretical object is the gap between a society's *coordination demand* $D$ (how much has to be coordinated:18people, territory, settlements, hierarchy of settlements) and its *coordination capacity* $K$ (administrative19institutions, standardized rules and money, communication infrastructure, record-keeping). The *coordination20pressure* is $P = z(\log D) - z(K)$. The Social Compression Hypothesis (SCH) says that institutions are compression21devices in the ordinary (not Shannon) sense: they lower the communication and verification cost per coordinated22task, which lets a society sustain larger $D$ at bounded cost, and that sustained high $P$ precedes instability.2324| ID | Testable statement (EMPIRICAL unless noted) | Datasets | Pre-specified falsifier |25|---|---|---|---|26| **H1** | Polities/societies with higher $K$ have larger scale ($\log$ population, $\log$ territory, settlement hierarchy), conditional on era, region, subsistence/production infrastructure and military technology. Direction of causality is **not** claimed. | Seshat (polity), D-PLACE EA (society), V-Dem+Maddison (country-year) | The partial association of $K$ with $\log D$ is not positive (95% CI includes 0 or is negative) in the primary Seshat specification, **or** its sign is not stable across the three datasets. |27| **H2** | Higher *pre-crisis* $P$ predicts more severe / more likely subsequent instability than conventional baselines. | Seshat CrisisDB (severity among crisis cases), Seshat polity duration (hazard), V-Dem (onset panel) | $\beta_P$ not positive at the Holm-adjusted 5% level **or** the $P$ model does not beat every baseline in out-of-sample log-likelihood (forward-chaining and leave-region-out). |28| **H3** (SIMULATED) | In agent-based experiments, shared protocols / standardized records / administration lower the number of messages and bytes per successfully coordinated task relative to no-institution baselines, at the price of fixed overhead. | Simulation (Phase 5) | Institutional regimes do not reduce cost-per-success in the majority of the factorial cells, or only do so in cells with implausible parameters. |29| **H4** | The relation between $P$ and instability is non-linear (threshold). | Seshat, V-Dem | A hinge/spline model does **not** improve out-of-sample log-likelihood over the linear-in-$P$ model in $\ge 80\%$ of folds. If so, H4 is rejected and only the smooth form is reported. |30| **H5** | H1/H2 estimates generalise across held-out macro-regions and eras. | Seshat, V-Dem | In any held-out macro-region or era block, the sign of $\beta_P$ reverses with a CI excluding the pooled estimate → the domain of the claim is narrowed to the blocks where it holds, and this is stated in the abstract. |3132Competing explanations recorded as covariates or rival models (never dropped): military competition (warfare33technology index), environmental/epidemic shock (CrisisDB epidemic flag used as covariate, never as outcome),34geography (region, UTM zone/latitude band), resource base (agricultural/production infrastructure index, subsistence35type in EA), inequality/stratification (EA class differentiation; Seshat has no direct measure → stated limitation),36coercion (professional soldiers/military officers), elite conflict (not directly measurable in Polaris → limitation),37cultural diffusion and non-independence (region/NGA random effects, language family in EA), selection (CrisisDB is38a crisis-selected sample → §3.2).3940---4142## 2. Units, samples and exclusions (fixed before analysis)4344### 2.1 Historical panel A — Seshat Polaris-202645- Unit: **polity** (`polity_id`, 626 in the `Polities` sheet). Reason: 24,401 of 26,033 social-complexity facts carry no46 `year_from`; they describe the whole polity period. A polity-year panel would be an interpolation that the data do not47 support. Within-polity dated facts (1,632) are used only (i) to build "pre-crisis" versions of $D$ when a dated value48 exists before the crisis start, and (ii) for descriptive trajectories.49- Time stamp of a polity: midpoint $(start\_year + end\_year)/2$; era blocks: $<-1000$, $-1000$ to $0$, $1$–$1000$,50 $1001$–$1500$, $1501$–$1800$, $>1800$ (CE). Region: `seshat_region` (42) nested in 10 macro-regions defined in51 `configs/regions.yaml` (fixed before analysis; continents/sub-continents).52- Exclusions: a polity enters a $P$-based model only if it has at least one coded $D$ scale variable53 (population or territory) and at least 50% of the $K$ indicators coded (non-missing after §4 rules). Polities coded54 `quasi-polity` in `polity_degree_of_centralization` are kept but flagged; a sensitivity analysis drops them.55- Duplicate entities: `polity_id` is unique in `Polities`; chained polities (e.g. `cn_qing_dyn_1/2`) are separate units56 and share an NGA random effect.5758### 2.2 Ethnographic cross-section B — D-PLACE Ethnographic Atlas v3.359- Unit: society (1,291 in `societies.csv`). Variables: EA031 (mean size of local communities), EA202 (population size),60 EA030 (settlement pattern) as $D$; EA033 (jurisdictional hierarchy beyond the local community), EA032 (local61 jurisdictional hierarchy), EA072/EA073 (political succession), EA090 (political integration) as $K$-side political62 organisation; EA066/EA067/EA068 (class/caste) as inequality covariates; EA005/EA028/EA042 subsistence covariates.63 No instability outcome exists in the EA → **B is used for H1 only.**64- Galton's problem: language family (Glottocode → family via Glottolog mapping included in D-PLACE), `region`, and65 spatial coordinates (Conley-type spatial HAC standard errors with 500 km and 1,000 km cut-offs).6667### 2.3 Modern panel C — V-Dem v16 + Maddison 202368- Unit: country-year, 1789–2025 (V-Dem). Country identifiers: V-Dem `country_id`/`country_text_id`; Maddison69 `countrycode` (ISO3) crosswalk by ISO3 — documented in `data/metadata/crosswalk_vdem_maddison.csv`; unmatched70 countries stay in V-Dem with population from `e_pop` only if its codebook source is Maddison-compatible (to be71 verified in the data dictionary; otherwise excluded).72- Exclusions: country-years with missing population; years before the first year a country's $K$ indicators are coded.73- **No link between C and A.** The two panels answer the same hypotheses separately.7475---7677## 3. Variables (names are the original source names)7879### 3.1 Coordination demand $D$ (strictly positive scale measures, logged)80Seshat: `polity_population` (persons), `polity_territory` (km²), `population_of_the_largest_settlement` (persons),81`settlement_hierarchy` (levels). Primary $D$ = first principal component of the available logged scale variables82**fit on the full Seshat sample without any outcome information**; secondary $D$ = `log polity_population` alone.83Range values (`value_from` ≠ `value_to`) → geometric midpoint, with the bounds carried as a measurement-error84sensitivity (§6). Modern: `pop` (Maddison, thousands) and `e_area` (V-Dem, km², 1886+) → primary $D$ = log population;85secondary adds log area when available. EA: EA031/EA202 ordinal categories → midpoints of the category ranges on a log scale.8687### 3.2 Coordination capacity $K$ (indicator sets are disjoint from $D$ and from $Y$)88Seshat, four blocks, each the share of indicators coded present (P or IP = 1; A or IA = 0; see §4):891. **Administration**: `full_time_bureaucrat`, `specialized_government_building`, `merit_promotion`,90 `examination_system`, `administrative_level` (levels; rescaled 0–1 by rank within the sample).912. **Standard rules and money**: `formal_legal_code`, `judge`, `court`, `professional_lawyer`, `indigenous_coin`,92 `foreign_coin`, `paper_currency`, `token`, `precious_metal`, `article`; measurement systems93 (`length_`, `area_`, `volume_`, `weight_`, `time_measurement_system`) enter only the latent model because they are94 coded for ≤ 139 polities.953. **Communication infrastructure**: `courier`, `postal_station`, `general_postal_service`, `road`, `bridge`, `canal`, `port`.964. **Record-keeping / information**: `written_record`, `script`, `phonetic_alphabetic_writing`, `non_phonetic_writing`,97 `lists_tables_and_classification`, `calendar`, `practical_literature`, `nonwritten_record`, `mnemonic_device`.9899Primary $K$ = unweighted mean of the four block scores (each block requires ≥ 50% of its indicators coded; a polity needs100≥ 3 blocks). Alternative $K$s (reported side by side, pre-specified): (a) 2-parameter logistic IRT latent score estimated101on all indicators of the four blocks; (b) first principal component of block scores; (c) block-by-block.102Explicitly **excluded** from $K$: military (`professional_soldier`, `professional_military_officer`, `military_level`,103Warfare sheet), religion (`professional_priesthood`, `religious_level`, Religion sheet), production infrastructure104(`irrigation_system`, `drinking_water_supply_system`, `market`, `food_storage_site`, `mines_or_quarry`), buildings,105literature genres other than practical literature. These form the rival-explanation covariates (§3.4).106Modern $K$ (V-Dem, none of them a democracy index): `v2clrspct` (rigorous and impartial public administration),107`v2stcritrecadm` (meritocratic appointment criteria), `v2cltrnslw` (transparent laws with predictable enforcement),108`v2stfisccap` (fiscal capacity source), `v2svstterr` (state authority over territory), `v2elembcap` (EMB capacity).109Primary = mean of standardized items with ≥ 4 coded; alternative = first PC. `v2x_polyarchy` is a **covariate**, never a110$K$ item.111EA $K$-side: EA033 levels (0–4) as the political-integration measure (ordinal, not logged).112113### 3.3 Coordination pressure $P$114$P = z(\log D) - z(K)$, standardization within dataset (Seshat: over all polities with both; modern: over all115country-years). $K$ is bounded and can be 0, so the $\log K$ form of the mandate is **not** used (documented deviation,116D-004). Sensitivities: standardize within era; within macro-region; use ranks.117118### 3.4 Covariates $X$ (rival explanations)119Seshat: warfare technology index (share of Warfare sheet indicators present), production infrastructure index,120moralizing-religion indicator (`moralizing_supernatural_punishment_and_reward`), `polity_degree_of_centralization`,121era block, macro-region, `polity_utm_zone`. Modern: log GDP per capita (Maddison `gdppc`), `v2x_polyarchy`,122year fixed effects, country fixed effects (within specifications), `e_regionpol_6C`, prior-instability indicator123(persistence). EA: subsistence (EA042), agriculture intensity (EA028), class differentiation (EA066), region, language family.124125### 3.5 Outcomes $Y$ (coded independently of $D$ and $K$)126- **Y1 (Seshat, primary for H2-A)** — crisis severity in CrisisDB "Crisis consequences": count of consequences coded127 P or IP among the 13 instability items {`decline`, `collapse`, `downward_mobility`, `extermination`, `uprising`,128 `revolution`, `successful_revolution`, `civil_war`, `century_plus`, `fragmentation`, `capital`, `conquest`,129 `assassination`, `depose`}, divided by the number of items coded P/IP/A/IA (SU, DIS, blank = missing, never zero).130 Binary version Y1b = any of {`collapse`, `fragmentation`, `century_plus`, `conquest`, `civil_war`} coded P/IP, defined131 only when all five are coded. `epidemic` is a covariate. Reform items (`constitution`, `labor`, `unfree_labor`,132 `suffrage`, `public_goods`, `religion`) are a separate exploratory outcome.133 **Selection caveat fixed in advance**: CrisisDB records exist for 150 of 626 polities and are a crisis-selected134 sample; the absence of a record is *not* "no crisis". Y1 therefore tests whether $P$ predicts *how badly* a crisis135 unfolds, conditional on a crisis being recorded, not crisis occurrence. Crisis records for the same polity (16136 duplicates) are separate crisis episodes; models cluster by polity.137- **Y2 (Seshat, secondary)** — polity duration $end\_year - start\_year$ with event = polity end; right-censored if138 `end_year` ≥ 1900 (coding boundary of the databank). Cox proportional hazards stratified by macro-region.139- **Y3 (Seshat, secondary)** — discontinuous succession: for polity $i$, the successor's140 `polity_relationship_to_preceding_entity` ∉ {continuity, continuation}; missing when no successor is in the databank.141- **Y4–Y6 (modern)** — onset within $(t, t+h]$, $h = 5$: Y4 civil-war onset (`e_civil_war` 0→1), Y5 coup142 (`e_coups` ≥ 1), Y6 regime end by non-constitutional means (`v2regendtype` categories to be read from the codebook and143 fixed in the data dictionary **before** estimation; documented in `research/measurement.md`). **Primary modern144 outcome** = composite Y456 = any of Y4–Y6 within the window. Persistence baseline uses the same outcome lagged.145146### 3.6 Temporal precedence (lag)147Seshat Y1: predictors use only facts whose `year_to` < crisis `year_from` **or** undated whole-polity facts148(unavoidable; flagged). A "strict" sensitivity uses dated-only facts (expected small N). Modern: $P_{it}$ with149outcomes in $(t, t+5]$; no contemporaneous terms; the persistence baseline uses $Y_{i,t-5..t}$.150151---152153## 4. Missing data, uncertainty, coding rules154- Seshat codes: `P`=1, `IP`=1 (flag inferred), `A`=0, `IA`=0, `A~P`/`P~A` = 0.5 (disputed; sensitivity: drop), `SU`,155 `U`, `unknown`, blank = **missing**. Missing is never zero. No forward-filling across polities or time156 (the Beheim et al. 2021 critique).157- Numeric ranges: midpoint (geometric for logged variables) with [`value_from`, `value_to`] as bounds; `is_uncertain`158 and `is_disputed` rows are flagged; sensitivity re-estimates with lower and upper bounds (two extra runs) and with159 flagged rows dropped.160- Dates: `year_from`/`year_to` kept as intervals; crisis start = `year_from`; no event is interpolated.161- Indices: an index is missing when fewer than 50% of its indicators are coded. Latent IRT model handles item162 missingness by likelihood (no imputation of the latent score for polities with < 3 coded items).163- Outcomes are never imputed.164165---166167## 5. Models (estimation commands live in `src/wp35/econometrics/`; seeds fixed at 20261101)168169### 5.1 H1170- Seshat: $\log D_i = a + b\,K_i + \gamma' X_i + \mu_{region} + \tau_{era} + e_i$, OLS, cluster-robust SE by171 `seshat_region`; also region random effects. Report $b$ with 95% CI.172- EA: ordinal logit of EA033 on $\log$ community size (EA031 midpoint) and covariates; spatial HAC SE; language-family173 clustered SE.174- Modern: within-country regression of $\log pop$ on $K$ with year FE (association only).175176### 5.2 H2 — Seshat Y1 (primary A)177Fractional logit (quasi-binomial) of Y1 on $P$ (pre-crisis) with macro-region FE and era block FE, SE clustered by178`seshat_region`. Binary Y1b: logit. Baselines fitted with identical samples: (B0) intercept + FE; (B1) population179only; (B2) $K$ only; (B3) Turchin-style single complexity PC1 (all SC indicators) ; (B4) persistence (prior crisis in the180same NGA). Model comparison: leave-one-region-out log-likelihood and AUC (Y1b), 10×5-fold CV; a model "beats" a181baseline if its mean OOS log-lik is higher in ≥ 80% of folds.182183### 5.3 H2 — Seshat Y2/Y3 (secondary)184Cox PH of duration on $P$, $X$, stratified by macro-region, Efron ties, robust SE by `seshat_region`; Schoenfeld test;185logistic for Y3.186187### 5.4 H2 — modern panel (primary C)188$\Pr(Y_{i,t+1..t+5}=1) = \Lambda(\alpha + \beta P_{it} + \gamma' X_{it} + \mu_i + \tau_t)$. Pooled logit with region and189year FE and country-clustered SE (primary, because country FE drop never-event countries), conditional (fixed-effects)190logit as secondary, linear probability with two-way FE as tertiary. Cubic polynomial of time-since-last-event (Carter &191Signorino 2010). Forward-chaining: train ≤ 1945 → test 1946–1975; train ≤ 1975 → test 1976–2005; train ≤ 2005 → test 2006+.192Leave-one-region-out on `e_regionpol_6C`. Metrics: OOS log-lik, AUC, Brier, calibration slope. Negative controls:193(i) outcome *before* $t$ regressed on $P_{t}$ with the same spec must not show a *stronger* association than the194forward spec; (ii) placebo $P$ built from the production/religion indicators excluded from $K$.195196### 5.5 H4197Linear-in-$P$ vs. hinge ($P$ and $(P-c)_+$, $c$ chosen by profile likelihood on the training fold only) vs. natural198cubic spline (3 df). Decision rule in §1.199200### 5.6 H5201Report $\beta_P$ per held-out macro-region and era block; a claim is retained only for blocks where the CI excludes zero202in the pooled direction; otherwise the domain statement is narrowed.203204### 5.7 Multiple testing205Three primary tests: H1-Seshat ($b$), H2-A ($\beta_P$ on Y1), H2-C ($\beta_P$ on Y456). Holm correction at family-wise2065%. All other estimates are labelled secondary or exploratory in every table.207208### 5.8 Power209Before estimation, a simulation-based minimum detectable effect (MDE, 80% power) will be computed for H2-A with the actual210$N$ and the actual distribution of $P$ (random outcome assignment, no real outcome values). Recorded in211`research/measurement.md`. If the MDE exceeds a standardized coefficient of 0.5, the Seshat H2 test is declared212*underpowered a priori* in the paper, whatever its result.213214---215216## 6. Robustness set (all pre-specified, all reported even when unfavorable)2171. $D$ alternatives (population only; territory only; PC1). 2. $K$ alternatives (mean of blocks; IRT; PC1; each block).2183. Standardization window (whole sample; within era; within macro-region). 4. Bounds of ranges (lower/upper).2195. Drop inferred codes (IP/IA → missing). 6. Drop quasi-polities. 7. Drop polities whose crisis starts within 25 years of220`start_year` (predictors may post-date). 8. Equinox-2020 release instead of Polaris-2026. 9. Modern: $h \in \{1, 3, 10\}$;221component outcomes; country FE logit; drop years < 1900. 10. Oster (2019) / Cinelli–Hazlett (2020) sensitivity to222unobservables for the primary coefficients.223224## 7. What would make us abandon the universal framing225If H2 fails in both A and C, the paper reports the SCH as **not supported** as a predictor of instability and keeps only226what survived (possibly H1 as an association and H3 as a SIMULATED result), with the title and abstract adjusted227accordingly. If H2 holds in one panel only, the claim is restricted to that panel's domain (H5).228